Geochemistry Geophysics Geosystems
3
G
Article Volume 7, Number 7 7 July 2006 Q07003, doi:10.1029/2005GC001120
AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society
ISSN: 1525-2027
Thermal and compositional structure of the subcontinental lithospheric mantle: Derivation from shear wave seismic tomography Tara J. Deen GEMOC ARC Key Centre, Department of Earth and Planetary Sciences, Macquarie University, North Ryde, NSW 2109, Australia Now at British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 0ET, UK (
[email protected])
W. L. Griffin GEMOC ARC Key Centre, Department of Earth and Planetary Sciences, Macquarie University, North Ryde, NSW 2109, Australia CSIRO Exploration and Mining, North Ryde, NSW 2113, Australia
G. Begg GEMOC ARC Key Centre, Department of Earth and Planetary Sciences, Macquarie University, North Ryde, NSW 2109, Australia WMC Resources, 191 Great Eastern Highway, Belmont, WA 6104, Australia Now at BHP Billiton Minerals Exploration, P.O. Box 91, Belmont, WA 6984, Australia
Suzanne Y. O’Reilly and L. M. Natapov GEMOC ARC Key Centre, Department of Earth and Planetary Sciences, Macquarie University, North Ryde, NSW 2109, Australia
J. Hronsky GEMOC ARC Key Centre, Department of Earth and Planetary Sciences, Macquarie University, North Ryde, NSW 2109, Australia WMC Resources, 191 Great Eastern Highway, Belmont, WA 6104, Australia Now at BHP Billiton Minerals Exploration, P.O. Box 91, Belmont, WA 6984, Australia
[1] Seismic tomography can provide unique information on the structure of the subcontinental lithospheric mantle (SCLM), but seismic velocity reflects both temperature and composition. We present a methodology for evaluating and isolating the relative contributions of these effects, which produces maps of regional geotherm and broad compositional constraints on the SCLM from the inversion of shear wave (Vs) seismic tomography. This approach uses model geotherms quantized in steps of 2.5 mW/m2 and three mantle compositions corresponding to typical Archean, Proterozoic, and Phanerozoic SCLM. Starting from an assumed composition for a volume of SCLM, lithospheric density at surface pressure and temperature is calculated for each geotherm at each point; the optimum geotherm is taken as the one yielding a density closest to the mean value derived from mantle xenoliths (3.31 g/cm3), since density varies with composition. Results requiring densities or geotherms outside the known natural range of these parameters worldwide require the choice of a different mantle composition. This technique, applied iteratively to a 275 km 275 km Vs model developed by S. Grand (University of Texas, Austin), results in maps of the geotherm and regional density, which allow interpretation of SCLM composition within broad limits. These results can Copyright 2006 by the American Geophysical Union
1 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
then be compared with local (paleo)geotherms and data for mantle composition, derived from xenolith suites. Application of this technique to the SCLM beneath Africa, Siberia, and North America shows good correlation with regional geological features, xenolith data, and other geophysical data. Components: 9282 words, 20 figures, 1 table. Keywords: seismic tomography; continental lithosphere; mantle composition; geotherms; lithosphere density. Index Terms: 3225 Mathematical Geophysics: Numerical approximations and analysis (4260); 3621 Mineralogy and Petrology: Mantle processes (1038); 7218 Seismology: Lithosphere (1236). Received 15 August 2005; Revised 1 January 2006; Accepted 23 March 2006; Published 7 July 2006. Deen, T. J., W. L. Griffin, G. Begg, S. Y. O’Reilly, L. M. Natapov, and J. Hronsky (2006), Thermal and compositional structure of the subcontinental lithospheric mantle: Derivation from shear wave seismic tomography, Geochem. Geophys. Geosyst., 7, Q07003, doi:10.1029/2005GC001120.
1. Introduction [ 2 ] The subcontinental lithospheric mantle (SCLM) is the lower part of the continental plate; it moves with the plate and carries a geochemical, thermal and chronological record of large-scale tectonic events that have shaped the Earth’s crust [O’Reilly et al., 2001]. Studies of mantle xenoliths and xenocrysts brought to the surface by volcanic rocks have shown that the composition of SCLM is broadly related to its tectonothermal age, as defined by the time of the last major tectonothermal events in the overlying crust [Griffin et al., 1999a, 2003b; and references therein]. We will use the tectonothermal classification of Griffin et al. [1999a] (modified from Janse [1994]): Archons experienced their last major tectonothermal event >2.5 Ga ago; Protons were formed or modified from 2.5–1.0 Ga, and Tectons since 1.0 Ga. The SCLM beneath Archons is generally strongly depleted in basaltic components (Ca, Al, Fe, etc.), while SCLM beneath Phanerozoic mobile belts (Tectons) is relatively fertile in terms of these components; SCLM beneath Proterozoic cratons and mobile belts (Protons) is generally intermediate between these two end-members. [3] While xenolith data provide a picture of mantle composition at individual localities in specific time slices, the limited geographic and temporal distribution of xenolith suites does not allow mapping of the large-scale, three-dimensional compositional structure of the SCLM. To extend our understanding of lithosphere composition and structure, we must turn to remote-sensing techniques, based on geophysical methods. The most spatially extensive data set available for directly imaging the Earth’s interior is global seismic tomography.
[4] Global tomographic studies show that continental ‘‘roots’’ with high seismic velocities extend to depths of 150–300 km under Archaean cratons [e.g., Jordan, 1988; Grand, 2002; Gung et al., 2003], while younger areas typically show thinner ‘‘roots’’, with somewhat lower velocities and extending to lesser depths. Seismic tomography is typically interpreted in terms of thermal variations. However, calculations of seismic velocities for mantle-derived xenoliths show that variations in mantle composition can account for as much as 25% of the observed velocity range [Griffin et al., 1999a; O’Reilly et al., 2001; Poudjom Djomani et al., 2001; James et al., 2003], with higher velocities corresponding to increasing degrees of depletion in basaltic components. Before seismic tomography can be used to map mantle composition, it is necessary to constrain the relative effects of temperature and composition, and relate both directly to variations in seismic wave speed. We will here show that by using constraints derived from a knowledge of the range of mantle compositions and the range of xenolith-derived geotherms, it is possible to invert tomography to obtain regional-scale maps of both the presentday geotherm and, within broad limits, mean mantle composition for the SCLM within the depth range 100–175 km (average depth of 140 km). [5] We use a variety of data sources to achieve this. Mantle-derived xenoliths provide information about the composition and thermal state of the mantle, and high-pressure, high-temperature experiments [e.g., Carmichael, 1984] provide the basis for calculations of the physical properties of mantle rocks at high pressures and temperatures. The known range of density in different mantle rock types provides constraints for relating exper2 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
imentally derived values to xenolith-derived compositional and thermal information, and thus to seismic velocities.
2. Rationale and Methodology [6] Geochemical studies have constrained the composition and thermal state (at the time of eruption) of limited volumes of the mantle which have been sampled as xenoliths or xenocrysts [e.g., Griffin et al., 1999a; O’Reilly and Griffin, 1996, 2006]. This study addresses the challenge of extending that knowledge to areas of the SCLM for which xenolith information is not available. A number of approaches have been developed in an attempt to solve the challenge of increasing the accuracy of estimates of lithospheric mantle composition. Deschamps et al. [2001] performed inversions of gravity data and seismic tomography models to develop a scaling factor which relates density anomalies to relative S wave velocity anomalies. They noted that there is a negative correlation between seismic velocities and gravity below continents to a depth of 200 km, as would be predicted from xenolith data. [7] Murikami and Yoshioka [2001] assumed a pyrolite composition for the mantle and determined values of bulk and shear moduli with respect to pressure, temperature and the Gru¨neisen-Anderson parameter for mineral composites of pyrolite composition. Their model assumed that Mg# (=100Mg/ (Mg + Fe)) is 89 for all minerals. However, in reality this ratio in olivine (the dominant mantle mineral) varies from 95 in highly depleted mantle rocks, to 88–89 in fertile mantle. This variation has a significant effect on the density and seismic velocity of the mantle. Using a forward modeling technique, Poudjom Djomani et al. [2001] calculated mean densities at 20C for Primitive Mantle (3.39 Mg m3), and for mean SCLM of Archaean (3.31 ± 0.16 Mg m 3 ), Proterozoic (3.35 ± 0.02 Mg m3) and Phanerozoic (3.36 Mg m3) tectonothermal ages. These values were derived by combining the mean modal and mineral compositions of mantle-derived xenoliths from areas of different tectonothermal history with high-pressure and high-temperature experimental data on the physical parameters of the constituent minerals. [8] Taking this approach a step further, we use lithospheric thermodynamics and the logarithmic equation of state (LES) [Poirier and Tarantola, 1998] to relate seismic velocity at depth to lithospheric geotherms and density constraints, to de-
10.1029/2005GC001120
velop estimates of mantle composition and geotherm for a given depth range. [9] The LES relates the density, pressure and rate of change of the bulk modulus with respect to the density at ambient pressure. It does not compensate for thermal effects, which must be done separately. We chose to use the logarithmic equation of state [Poirier and Tarantola, 1998] rather than the more commonly used Birch-Murnaghan (B-M) equation because it is based on Hencky logarithmic strain, rather than Eulerian strain. While Hencky strain is approximately equivalent to Eulerian strain for small strains, it is more robust than Eulerian strain for larger strain values and allows for a more reliable extrapolation to pressures outside of the experimental range [Poirier and Tarantola, 1998]. The LES has a simpler expression, and while the differences in results between the two equations are minimal at low strains, experiments suggest that the LES is more accurate than the B-M at higher strains [Poirier and Tarantola, 1998]. Initial testing also showed that the LES appeared to give more realistic results, compared to xenolith data, than the B-M formulation. [10] The procedure is summarized in Figure 1. The first step in this inversion is to calculate the in situ lithospheric density r(P,T) for an assumed mantle composition. Taking the lithospheric shear wave velocity v(P,T) and determining the shear modulus at @m @m depth (m(P,T) = m(0,298K)+ (@T T) + (@P P)) for an assumed composition and geotherm, we determine the value of the density of the lithosphere at depth r(P,T), using the relationship rðP;T Þ ¼
mðP;T Þ v2sðP;T Þ
With the in situ density calculated, the pressure effects are then removed using values calculated for the bulk modulus at ambient pressure temperature (K(0,T) = and lithospheric @Kð0;298KÞ K0 + T ) and the pressure derivative of @T @Kð0;T Þ ) for the assumed the bulk modulus (K00 = @P lithospheric composition by using a Newton’s approximation to solve the logarithmic equation of state [Poirier and Tarantola, 1998] for r(0,T): 2 !2 3 0 rðP;T Þ rðP;T Þ rðP;T Þ K 2 0 4ln 5 ln P ¼ Kð0;T Þ þ rð0;T Þ rð0;T Þ rð0;T Þ 2
A further step is required to bring the density to room temperature values. The conversion from 3 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Figure 1. Flowchart indicating the progress of the model inversion, to remove temperature and pressure effects from density.
lithospheric temperature is calculated using the relationship rð0;298K Þ ¼
rð0;T Þ RT 1 að0;T Þ dT
where a0, a1, a2 and a3 are the coefficients obtained from experimental results. The thermal expansion coefficient is determined from experimental data and is dependent on composition and temperature.
0
where a(0,T) is the experimentally derived volumetric thermal expansion coefficient at ambient pressure (1 bar) and temperature T, given by [Fei, 1995] að0;T Þ ¼ a0 þ a1 T þ a2 T 1 þ a3 T 2 þ . . .
[11] This approach requires a geotherm, in order to determine the necessary temperature corrections. In addition, the application of stepped geotherms allows an assessment of the accuracy of the removal of temperature and pressure effects from the density during the modeling procedure. The geotherms used in this work are the classic model conductive geotherms of Pollack and Chapman [1977], parameterized in 2.5 mW/m2 steps between 4 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Figure 2. Parameterized reference geotherms derived from the work of Pollack and Chapman [1977], labeled according to corresponding surface heat flow in m/Wm2. The southeastern Australia advective geotherm (SEA) is derived from the geothermobarometry of xenoliths in basalts [O’Reilly and Griffin, 1985].
30 mW/m2 and 60 mW/m2, and an advective empirical xenolith-derived geotherm for SouthEast Australia (SEA) (O’Reilly and Griffin [1985] and updates by O’Reilly et al. [1997], Xu et al. [1998], Yu et al. [2003], and Gaul et al. [2003]) which is appropriate for many Phanerozoic areas, particularly those with Tertiary to Recent volcanism (Figure 2). Corrections were not made for local variations in Moho depth; however, this work focuses on the lithosphere between 100 and 150 km where these effects are minimal. Such empirical xenolith-based geotherms are independent of uncertainties in the heat production and thermal conductivity of different lithospheric layers. They are based on geothermobarometric determinations on a set of xenolith or xenocryst samples from a given region (see summary of methods by O’Reilly and Griffin [2006]). These represent the geotherm at the time of eruption of the host volcanic rock. The model conductive geotherms reflect particular assumptions about the distribution of heat sources in the crust and mantle, but they are useful because they approximate the P-T arrays derived from xenolith studies, and can be parameterized. There is a broad worldwide correlation between tectonothermal age, SCLM composition and geotherm; most Archon areas are characterized by low (30– 40 mW/m2) geotherms, and most Proton areas by higher conductive geotherms (40 – 55 mW/m2) [O’Reilly and Griffin, 1996, 2006; O’Reilly et al., 2001].
[12] Our approach requires initial assumptions to be made about composition for a given area, due to the compositional dependency of the bulk and shear moduli and the thermal expansion coefficient. Because there are large differences in mean composition, and hence in density, between SCLM of different tectonothermal age [Poudjom Djomani et al., 2001; Griffin et al., 1999a; O’Reilly et al., 2001; O’Reilly and Griffin, 1996, 2006], the choice of composition will strongly influence the resulting estimates of density and geotherm for each point. To reduce the effect of bias derived from initial compositional estimates, we have derived maps of the geotherm and mean density (for the given depth slice) assuming three different compositions, corresponding to mean Archean, Proterozoic and Phanerozoic SCLM (Table 1). The joint inversion of both density and geotherm is possible because (1) both parameters vary within relatively narrow limits in nature and (2) we are optimizing only for a choice of one of three generalized compositions and using a series of geotherms stepped at relatively wide intervals. The inversion of the geotherm thus is optimized with respect to a narrow range of realistic densities, defined by the forward modeling from xenolith data [Poudjom Djomani et al., 2001]. [13] Taking the initial composition, the inversion calculates the density across a range of geotherms (Figure 2); the optimum geotherm is accepted as that which yields the most realistic density, defined 5 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Table 1. Mineral Composition, Density, Predicted Surface Vs Seismic Velocity, Bulk Modulus K, Shear Modulus m, and Temperature and Moduli Pressure Derivative Variation for Subcontinental Mantle Lithosphere of Varying Tectonothermal Agea
Rock Type Archean Lherz Fo93 Archean Lherz Fo92 Archean Harz Fo93 Archean Harz Fo92 Phanerozoic Fo 90 Proterozoic Fo91 Proterozoic Fo92 Primitive mantle a
% % % % % oliv enst cpx gnt spin 0.69 0.69 0.65 0.65 0.67 0.60 0.65 0.70 0.70 0.57
0.25 0.25 0.30 0.30 0.16 0.17 0.23 0.18 0.18 0.13
0.02 0.02 0.00 0.00 0.09 0.11 0.10 0.06 0.06 0.12
0.04 0.04 0.05 0.05 0.09 0.12
0.00 0.00 0.00 0.00
0.02 0.07 0.00 0.07 0.00 0.18 0
Vs Bulk Shear Density Vp 200C 200C 200C Modulus K dK(0,T)/dP Modulus m(0,298) dm(P,T)/dP dm(P,T)/dT (YP) (YP) (YP) (YP,HPP) (HPP) 3.31 3.32 3.31 3.32 3.37 3.38 3.34 3.35 3.34 3.42
8.34 8.33 8.34 8.33 8.32 8.32 8.26 8.33 8.35 8.31
4.88 4.87 4.88 4.87 4.82 4.81 4.81 4.84 4.85 4.81
125.24 123.39 125.01 123.27 126.69 128.38 127.05 124.40 131.32
5.07 5.11 5.07 5.11 5.12
78.49 78.40 78.74 78.65 78.09 78.29
1.84 1.84 1.86 1.85 1.80 1.79
0.0125 0.0126 0.0125 0.0125 0.0123 0.0121
5.08 4.82
78.44 80.87
1.81 1.77
0.0125 0.0120
Figures compiled from Murikami and Yoshioka [2001], Poudjom Djomani et al. [2001] (YP), and Carmichael [1984] (HPP).
as the value closest to the mean of 3.31 g/cm3 [Poudjom Djomani et al., 2001]. In practice, only one of the model geotherms will typically give a density value in the range 2.9 – 3.32 g/cm 3 , corresponding to known major mantle rock types. Initially assuming an incorrect composition may result in realistic densities due to the density optimization process, while calculating a realistic density from the wrong composition forces the calculation of an unrealistic corresponding geotherm for the area. This ‘‘dual testing’’ approach reduces compositional bias by removing the need for an initial compositional estimate. Comparison of the mono-compositional maps of geotherm and density allows us to identify areas where, for instance, an Archean or a Phanerozoic mantle composition yields an unrealistic geotherm, and areas where a realistic geotherm requires an unrealistic mantle density. In a second iteration, we replace the mantle composition beneath selected polygons with the composition that has yielded a realistic value of both geotherm and density, and which is at least broadly consistent with the tectonothermal history of the crust. [14] The seismic data used in this work are derived from a global, 275 275 km block shear wave model [Grand, 2002], which was reprocessed by WMC Resources to interpolate a global, 100 100 km model. The original model was derived in the customary inverse manner as deviations from the Preliminary Reference Earth Model (PREM) [Dziewonski and Anderson, 1981]; the absolute values of seismic velocity required for this work were derived by adding the deviations to the PREM. The seismic data as presented here
(Figure 3) have a color scale which is the reverse of that normally used for presenting tomographic images; slower velocities are represented as blue to black, and faster velocities as red to white. This has been done to remove the a priori, but commonly misleading, assumption that high seismic velocities correlate with lower temperatures, which is implied by the traditional color scale. This choice of color scale also allows direct comparison of the seismic data with the density maps produced by the modeling. The seismic velocities used for the examples shown below were calculated for the depth slice
Figure 3. Distribution of shear wave seismic velocity (Vs) for the range 100 –175 km beneath Africa. Zones of relatively higher shear wave velocity are red and white, while zones of slower velocity are blue and black. Tomography reprocessed by WMC Resources (Perth) [from Grand, 2002]. 6 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Figure 4. (a) Upper Lithospheric Domains (ULDs) indicating the tectonothermal age of lithospheric volumes determined from mapping of geology and geophysics [Griffin et al., 2005], labeled with broad geographical regions referred to in the text. A, Archon; P, Proton; T, Tecton; P/A, Proton-reworked Archon; T/P, Tecton-reworked Proton. (b) Lower Lithospheric Domains (LLDs) of tectonothermal ages developed through analysis of the results summarized in Figures 5 – 7.
between 100 km and 175 km below the surface of the Earth (Figure 3). [15] The process summarized in Figure 1 was applied to each point interpolated from the seismic
model individually, with no reference to the values of surrounding points. This approach was used to partially compensate for any bias introduced into the model from the assumption of composition,
Figure 5. (left) Calculated density and (right) corresponding geotherm values in mW/m2 for the lithosphere beneath Africa, assuming a homogeneous Archon composition. The two results must always be considered in conjunction with each other (see text). Imposing an Archon composition results in realistic densities and geotherms within the Archon Cratons in the western part of Africa but cannot explain the velocity and temperature conditions in the north and east of the continent. 7 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Figure 6. (left) Calculated density and (right) corresponding geotherm values in mW/m2 for the lithosphere beneath Africa, assuming a homogeneous Proton composition.
and the bias each calculation may have on those points surrounding it.
3. Results 3.1. Africa [16] Figure 3 shows the Vs model for the 100– 175 km depth slice beneath Africa; Figure 4a shows a tectonic synthesis in which crustal domains are classified by the timing of the last major tectonothermal event to affect the crust. The major features of this map are (1) a series of Archean cratonic blocks (A), several of which were reworked to varying degrees in Proterozoic time (P/A), (2) Proterozoic mobile belts (P), an assemblage of Archean to Proterozoic crustal fragments swept together in Pan-African time, or otherwise modified since 1.0 Ga (T/P), and (3) narrow zones affected by post-Pan-African rifting (T). Evidence on mantle composition and on the local geotherm at the time of eruption is provided by xenolith and xenocryst data from a large number of geographically widespread kimberlites, mostly from southern Africa and Tanzania (Figure 3) [Griffin et al., 2003a, and references therein]. [17] In Figure 3 the West Africa Craton and Congo Craton show up as relatively homogeneous, highvelocity masses; the Kalahari Craton is more diffuse, with the highest velocities under the NE part in Zimbabwe. Large regions of moderate to high velocity are present at 100–175 km depth
west of the West African coastline [O’Reilly and Griffin, 2004]; the largest of these areas is the extension of the Congo Craton. The Sahara MetaCraton (also known as the Ghost Craton) in northern Africa is a collage of Proterozoic to Archean crustal fragments, assembled in Pan-African time; it shows up as a block with a range of relatively low Vs (yellow-orange tones). Areas of moderate Phanerozoic extension across central Africa and Cameroon, and the Damara Orogen in SW Africa, have still lower velocities (yellow-green tones). The East Africa Rift has very low seismic velocities in the 100–175 km depth range, shown as deep blue to black tones. A well-defined lowvelocity region beneath the Hoggar Swell extends to 400 km and in fact, the LES cannot be reasonably solved for Vs values in this region; melts or fluids may be responsible for the low Vs. The Vs anomaly may image a mantle hot spot responsible for the regional uplift and the Cenozoic volcanism of the Hoggar swell [O’Reilly and Griffin, 2004]. [18] Figures 5 to 7 show the single-composition maps of the SCLM at a depth of 100–175 km below Africa. The color scale of the density images is optimized to highlight features within the expected range of 2.9 g/cm3 to 3.32 g/cm3 for lithospheric densities. Densities shown as dark blue on the images are lower than expected from forward modeling (especially if they are matched by low geotherms), while densities shown in yellows are equivalent to the density of Phanerozoic (Tec8 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Figure 7. (left) Calculated density and (right) corresponding geotherm values in mW/m2 for the lithosphere beneath Africa, assuming a homogeneous Tecton composition.
ton) lithosphere. Reds on the images represent values higher than expected from forward modeling of xenolith data, especially if accompanied by higher geotherms. As the higher geotherms considered here approach the mantle adiabat [e.g., McKenzie and Bickle, 1988] at lithospheric depths, we have biased our inversion to default to the SEA geotherm in that circumstance. Consequently, the SEA geotherm (shown in red), indicates either that the geotherm is equivalent to the SEA geotherm (the case for young mantle), or that the adiabat has been reached at 140 km depth in that region. [19] For each of these maps, the calculated density of the SCLM is close to 3.3 g/cm3 for most of the continent (Figure 5), which is a result of the optimization process. Those areas which are closest to a density of 3.3 g/cm3 have the highest accuracy in the inversion results. Younger areas at plate margins such as the Atlantic Ocean generally give a density of 3.36–3.38 (shown in yellow), which is consistent with Tecton or Primitive mantle (Table 1). The Hoggar Swell and the East Africa Rift give densities well above 3.4 gm/cm3, outside the observed range for mantle peridotites for a realistic geotherm. This suggests that a partial melt fraction is present at this level in the mantle (100– 175 km). The presence of fluid will reduce the in situ shear wave velocities, which the inversion will interpret as corresponding to artificially high densities. While some of the temperature anomaly in the region could be explained by anelasticity, Karato [1993] indicated that if anelasticity were
not taken into account, a 5% Vs anomaly would result in the overestimation of DT by 300 K, for Qu 100. Venkataraman et al. [2004] performed joint P wave velocity perturbations and QP values for the East African Rift, taking into account anelasticity after Karato [1993], and found that the upper mantle temperatures (100 – 400 km depth, QP 80) are 140 – 280 K higher than ambient mantle temperatures. The Grand [2002] model indicates a temperature anomaly of >300 K due to a Vs anomaly of 2%. This indicates that there are further processes at work in the SCLM of the East Africa Rift than anelasticity. In addition, the mantle temperatures calculated by Venkataraman et al. [2004] and this work indicate that the EAR lithosphere is at a temperature that would permit partial melt. [20] Comparing the effect on the geotherm of forcing the density to 3.3 for a given composition allows the assessment of the likely composition of the region under investigation. An assumption that the entire African SCLM is Archon in composition (Figure 5) results in geologically reasonable geotherms of to 32.5–35 mW/m2 within most the Archean cratons; lower geotherms (30 mW/m2) are limited to a strip along the western side of the West Africa craton and small regions within the Congo Craton. However, even geotherms higher than 60 mW/m2 (resulting in temperatures equivalent to the mantle adiabat in this depth range; Figure 2) are insufficient to balance the anomalously high densities required by an Archon com9 of 20
Geochemistry Geophysics Geosystems
3
G
deen et al.: subcontinental lithospheric mantle
10.1029/2005GC001120
Figure 8. (left) Calculated density and (right) corresponding geotherm values in mW/m2 (right) for the lithosphere beneath Africa, using tectonothermal ages determined from Figures 5 – 7 and outlined in Figure 4b.
position on the eastern side of Africa, in the Sahara Meta-craton, and with increasing proximity to the East Africa Rift. [21] The assumption of a uniform Proton composition for the SCLM (Figure 6) provides more geologically reasonable results, particularly for the Sahara meta-craton. However, imposing a Proton composition on the Archean cratons results in modeled densities lower than those measured in xenoliths from such cratons [Griffin et al., 1999b; James et al., 2003]. These low densities cannot be compensated even by geotherms
